摘要
针对遗传算法特征选取技术,提出4种适应度函数构造方法,即,基于改进的距离判据、基于平均值方差比、基于Fisher判别准则以及基于最近邻分类法的适应度函数,并通过仿真实例对方法进行了验证。最后,将新方法应用于转子故障诊断,结果表明:笔者提出的遗传算法特征选择的4种适应度函数的正确有效性。
In this paper,aiming at feature selection based on genetic algorithm,four fitness functions are constructed,that is the function of improved distance criterion,of mean-variance ratio,of Fisher criterion and the most near neighbor classifying function.These methods are verified by a simulation example.Finally,The new methods are applied to diagnose the rotor faults,and the results show that the new methods are correct and effective.
出处
《机械科学与技术》
CSCD
北大核心
2011年第1期124-128,132,共6页
Mechanical Science and Technology for Aerospace Engineering
基金
国家自然科学基金项目(50705042)
航空科学基金项目(2007ZB52022)资助
关键词
特征选取
遗传算法
适应度函数
转子
故障诊断
feature selection
genetic algorithm
fitness function
rotor
fault diagnosis
